US Judge Orders OpenAI to Share 20 Million ChatGPT Chats in Copyright Case

A federal judge in New York has ordered OpenAI to hand over 20 million anonymized ChatGPT conversations as part of a major copyright lawsuit involving artificial intelligence. The decision marks an important step in a growing legal battle over how AI systems are trained and whether they unlawfully use copyrighted material.

On January 5, 2026, U.S. District Judge Sidney H. Stein upheld an earlier order issued by Magistrate Judge Ona T. Wang. The ruling requires OpenAI to provide a large set of user chat logs to plaintiffs, which include several major news organizations and content creators. These records must be anonymized and shared under strict protective rules to safeguard user privacy.

US Judge Orders OpenAI to Share 20 Million ChatGPT Chats in Copyright Case

OpenAI opposed the order, arguing that producing such a large dataset would be burdensome and could risk exposing sensitive user information. The company said the 20 million chats represent about 0.5 percent of all preserved ChatGPT logs and suggested a narrower approach. Instead of releasing the full dataset, OpenAI proposed searching only for conversations that directly referenced the plaintiffs’ copyrighted works.

Judge Stein rejected this argument. He noted that U.S. law does not require courts to force parties to use the “least burdensome” method of discovery. According to the judge, the anonymization measures and protective order are sufficient to address privacy concerns while allowing plaintiffs access to relevant evidence.

The dispute began in July 2025, when organizations such as The New York Times and the Chicago Tribune sought access to as many as 120 million ChatGPT conversations. Their goal was to determine whether ChatGPT reproduced copyrighted content that it may have been trained on. After negotiations, OpenAI offered a reduced dataset of 20 million chats. While the plaintiffs initially accepted this compromise, OpenAI later argued that most of the data would be irrelevant, a position the court ultimately rejected.

OpenAI also cited a past securities case from the Second Circuit that blocked the disclosure of wiretap recordings. Judge Stein dismissed the comparison, explaining that ChatGPT conversations involve voluntary user inputs owned by the company, not secretly recorded communications. Both judges emphasized that strong de-identification methods would protect users’ identities.

See Also: Rushing 5G Without Affordable Phones Could Strain Foreign Exchange Reserves, Warns Telecom Industry Chief

This ruling moves forward the case known as In re OpenAI, Inc. Copyright Infringement Litigation, which combines 16 lawsuits from publishers, authors, and other rights holders. These cases claim that AI companies used copyrighted works without permission to train large language models. Other technology firms, including Microsoft and Meta, have faced similar lawsuits, making this a closely watched legal issue.

Supporters of the decision argue that access to large datasets is necessary to evaluate the scope of potential copyright infringement and to challenge OpenAI’s defenses. Critics, however, worry that mass disclosure of chat logs—even anonymized—could weaken public trust in AI tools.

Legal experts say the ruling highlights the tension between privacy, corporate data practices, and transparency in AI development. As lawsuits against AI companies continue to increase, courts appear more willing to compel the release of extensive evidence. This case may set an important precedent for how courts handle user data and copyright issues in the age of generative artificial intelligence.